4 DS DIMMs at 1T: Memory Stress Testing

The DFI easily handles 2-2-2-7-1T timings at stock speed with 2 double-sided dimms, as do almost any of the current boards for AMD Socket 939 from NVIDIA, SiS, VIA, and ULi. However, what is truly unique is that the DFI LANParty UT RDX200 is the first board designed to run four double-sided 512MB or 1GB DIMMs at the fastest timings and a 1T Command Rate. In fact, AMD specifications tell us that this is impossible. To his credit, Oskar Wu and DFI found a way to accomplish the impossible.

Stable DDR400 Timings - 4 DIMMs
(4/4 DIMMs populated)
Clock Speed: 200MHz
CAS Latency: 2.0
RAS to CAS Delay: 2T
RAS Precharge: 7T*
Precharge Delay: 2T
Command Rate: 1T
*7T was determined by MemTest86 benchmarks to deliver the widest bandwidth with the ATI Rx480/482 chipset. While the board would operate at tRAS of 5T or lower, all benchmarks were run at 7T.

With the ability to run 4 DS DIMMs at 1T, we had the opportunity to finally show the impact of a drop to a 2T Command Rate on performance.

First, for those skeptical that this is possible, we captured screens from the freeware A64 Tweaker and CPU-Z showing operation of 4 DIMMs at 1T.


Click to enlarge.

All the benchmarks below were run with an FX-57 processor and are not directly comparable to our standard benchmarks which use an AMD 4000+ CPU.

To clearly demonstrate the impact of the normal 2T drop in Command Rate on performance, we ran Aquamark3 benchmarks with 2 DIMMs at 1T, 4 DIMMs at 2T, and the new option of 4 DIMMs at 1T.

2 DIMMs at 1T is normally the fastest available option.

You can see that doubling the memory, but being forced to drop to 2T Command Rate, lowers the A3 scores by over 1,000 points.

The surprise, however, is the new DFI option of 4DS DIMMs at 1T. Not only is there no performance loss - the scores are 2,000 points higher than 4DSx2T - but 4DS DIMMs at 1T outperforms 2x1T by almost 1,000 points. This makes 4DS DIMMs at 1T in the DFI to be the new performance leader at stock speeds.

For those who are thinking that surely there's a catch, the answer is "yes, there is". 1T operation is only possible to a frequency of about DDR406, which basically limits it to stock speeds. Above that speed, you must run 4 DS DIMMs at 2T as you normally must. However, at or near stock speeds, the 1T option is a brilliant development that delivers better performance.

Overclocking: DFI LANParty UT RDX200 Test Setup
Comments Locked

48 Comments

View All Comments

  • Skoad - Friday, October 21, 2005 - link

    Board just came in stock at newegg for $209+5 shipping.


    Also what psu was used in this test. I read somewhere that the board needs an 8pin connection from the psu and that very few psu's have this right now.

    I can't seem to find where I read this at atm.
  • Wesley Fink - Friday, October 21, 2005 - link

    We use the OCZ 520W as a standard bench PSU. It has both 4 and 8-pin 12V connectors. 8-pin slots will also work with 4-pin 12V plugs.
  • Zoomer - Wednesday, October 19, 2005 - link

    Will there be a value ATi board?

    I don't forsee myself getting a xfire solution, ever.
  • smaky - Thursday, October 20, 2005 - link

    Guess who is getting one this week? hiihihihiihih x850?
  • danidentity - Wednesday, October 19, 2005 - link

    You make no comment on the stability of this board, how is it?
  • cryptonomicon - Wednesday, October 19, 2005 - link

    Sorry, but I can't help but ask why the gaming performance graphs were not all done on the same graphics card. Initially I thought WOW THIS NEW ATI CHIPSET IS MAD FAST but then I see it was using the 7800 gtx while all the other boards got 6800 ultras... WTF?
    What the heck is going on? Was the scientific method forgotten or something? This is a let down.
  • rjm55 - Wednesday, October 19, 2005 - link

    I'm always amazed in reading AT Comments that those who complain loudest are the ones who don't even bother to READ the review. I may not always agree with Wesley's approach on a review, but I know his results are always documented clearly in the review, logical, and repeatable. In fact he is the reviewer at AT who is most careful to always document the components he tested with and the setups. Derek and Anand often leave you guessing how they tested and you have to ask to figure it out.

    If you had bothered to read the test setup you would have seen that the red bars are tests with the 6800 Ultra - the same used in every other compared board. The 7800GTX and Crossfire were BONUS results - for those who would be screaming "Why have you tested with the obsolete 6800 Ultra instead of the 7800GTX".

    Please READ before you scream so your rants aren't a total waste of time.
  • JarredWalton - Wednesday, October 19, 2005 - link

    Sorry to have to point this out, but assembling and testing a motherboard can take several days. The 7800GTX scores were there for reference, but 6800 Ultra was used as well (the red bar) to make scores comparable. It's not practical to go back and retest seven (or more) motherboards every time a new article needs to be published. If we don't include something like the 7800GTX, people wonder how that affects performance. Just look at the red bars for motherboard comparisons and the gold bar to see what a $500 (instead of $350) GPU will do. :)
  • cryptonomicon - Wednesday, October 19, 2005 - link

    nevermind, wesley made no error, just me :D the board using the 6800u in red is the direct comparison (which it said in the article, albeit not on the gaming performance page). must have been too late at night or i was just too dumb to see it! excellent article as always.
  • cryptonomicon - Wednesday, October 19, 2005 - link

    hats off to DFI, this board is wicked fast. i am curious though as to how it will sell given the high price.

Log in

Don't have an account? Sign up now